A Semantic-based Medical Image Fusion Approach

This is the code implementation of the paper "A Semantic-based Medical Image Fusion Approach"

Dependencies

Contents of requirements.txt:

pytorch==1.0.1
torchvision==0.2.2
scikit-image==0.14.1

Install dependencies by the following command(recommended to use conda):

conda install --yes --file requirements.txt

Prerequisite

Our data was downloaded from the Harvard Medical School website. The script for downloading data is as follows:

python3 download_data.py

Use the following code to generate a dataset partition used by pytorch.

python3 generate_path.py

All images are registered. Image examples:

Fusion task

We saved the parameters of the FW-Net model used in the paper, and you can directly predict the input image using the following code:

python3 predict.py ct_img_path mr_img_path

If you want to train your own model based on your dataset, the training code is visualized in train.ipynb, or you can execute python3 train.py directly.

The comparative methods used in this article can be found in the following links: GF, NSCT-PCDC, NSCT-RPCNN, LP-CNN and NSST-PAPCNN

Semantic loss evaluation task

The code of semantic loss network is in semantic_loss_metric.ipynb.

Reference